RAG, retrieval, factuality

BEIR, FActScore, RAGAS, Self-RAG, ARES, SAFE, SimpleQA, HalluLens.

Eight papers covering retrieval, RAG eval, and factuality scoring. BEIR is the retrieval baseline that survives. FActScore is the atomic-fact pattern that everything else borrows from. RAGAS, Self-RAG, and ARES wire those metrics into pipelines. SAFE is the search-augmented evaluator result that makes large-scale factuality scoring cheap. SimpleQA and HalluLens are the 2024 to 2025 work to read before claiming a hallucination rate.

AI Evals

The most useful, opinionated reference for AI evals.

AI Evals is a curated technical reference for evaluating large language model systems in production. The site covers error analysis, LLM-as-judge calibration, RAG evaluation, agentic eval, statistical rigor, and twenty task-specific eval playbooks. Every claim is cited.

Why evals matter
Your first eval in 30 minutes
LLM-as-Judge introduction
RAG eval in two layers
Trajectory eval for agents
Glossary
Papers